A method for the continuous recognition of gestures of a user of a handheld mobile terminal fitted with a motion sensor assembly, includes the steps of, for at least one motion sensor measuring signal of the motion sensor assembly, over a time window: The candidate gesture detection step (2) includes substeps of:
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A method for continuous recognition of gestures of a user of a handheld mobile device comprising a motion sensor assembly, the method including steps comprising, for at least one motion sensor measuring signal of the motion sensor assembly, over a time window: detecting a candidate gesture; and comparing said candidate gesture with a set of predetermined gestures; said candidate gesture detection step including substeps comprising: calculating a value of a parameter indicative of a symmetry of the candidate gesture with respect to at least one axis, indicating for the candidate gesture, the extent to which the handheld mobile device is returned to at least one of: an initial position, and an initial orientation; and calculating a value of a parameter indicative of an energy of the candidate gesture along at least one axis.
A method for recognizing gestures on a handheld mobile device with a motion sensor. The method continuously monitors sensor signals over a time window to detect gestures. It calculates two parameters: 1) a "symmetry" value indicating how well the gesture returns the device to its starting position or orientation, and 2) an "energy" value representing the gesture's intensity along at least one axis. These parameters help identify potential gestures by quantifying their shape and strength. The system then compares the detected candidate gesture with a set of stored, known gestures for recognition.
2. The method according to claim 1 , wherein said parameter indicative of the symmetry of the candidate gesture with respect to at least one axis depends on an integral of said signal.
In the gesture recognition method, the "symmetry" parameter, which indicates how closely a gesture returns the device to its initial position, is calculated using an integral of the motion sensor's signal over time. This integral captures the overall displacement and helps assess the symmetry of the movement. Therefore, the symmetry parameter depends on the accumulation of the signal over the time window.
3. The method according to claim 1 , wherein said parameter indicative of the symmetry of the candidate gesture with respect to at least one axis depends on an average of said signal.
In the gesture recognition method, the "symmetry" parameter, which indicates how closely a gesture returns the device to its initial position, is calculated using an average of the motion sensor's signal over time. This average smooths out the signal and provides a general indication of the device's position relative to its starting point, making the symmetry calculation more robust to noise. Therefore, the symmetry parameter depends on the mean of the signal over the time window.
4. The method according to claim 1 , wherein said parameter indicative of the energy of the candidate gesture along at least one axis depends on a variance of said signal.
In the gesture recognition method, the "energy" parameter, which represents the gesture's intensity, is calculated using the variance of the motion sensor's signal along at least one axis. The variance measures how much the signal deviates from its mean, capturing the dynamic range of the motion and reflecting the gesture's forcefulness. Therefore, the energy parameter depends on the spread of the signal over the time window.
5. The method according to claim 1 , wherein said candidate gesture detection step further comprises a substep of testing if the value of the parameter indicative of the symmetry of the candidate gesture with respect to at least one axis is less than or equal to a first threshold and if the value of the parameter indicative of the energy of the candidate gesture along said axis is greater than or equal to a second threshold.
In the gesture recognition method, after calculating the symmetry and energy parameters, the method checks if the symmetry value is below a certain threshold (first threshold) and if the energy value is above another threshold (second threshold). Only if both conditions are met is a candidate gesture considered valid. This filtering step helps eliminate false positives by ensuring that the gesture is both symmetrical enough and energetic enough.
6. The method according to claim 5 , wherein said candidate gesture detection step further includes a substep of testing if a value of a third parameter indicative of a dominant frequency of the signal along said axis is less than or equal to a third threshold.
In the gesture recognition method described previously, after checking symmetry and energy against thresholds, a third parameter, the "dominant frequency" of the sensor signal, is calculated. The method then tests if this dominant frequency is below a third threshold. The dominant frequency represents the primary oscillation rate of the gesture, and this filtering step helps to differentiate between different types of gestures based on their speed and rhythm.
7. The method according to claim 6 , further including a step of automatic adaptation of at least one of said first, second, and third thresholds to the user.
The gesture recognition method automatically adjusts at least one of the three thresholds (for symmetry, energy, and dominant frequency) to adapt to the specific user. This personalization allows the system to account for variations in how different users perform the same gesture, improving recognition accuracy over time. The adaptation occurs dynamically based on the user's actual gesture patterns.
8. The method according to claim 7 , further including a step of modifying at least one of said first, second and third thresholds when a detected candidate gesture includes a corresponding parameter of a value not within a range of percentage values of said corresponding threshold.
The gesture recognition method adapts the symmetry, energy, and dominant frequency thresholds by monitoring detected gestures. If a gesture's parameter value (e.g., symmetry) falls outside an acceptable percentage range of the corresponding threshold, that threshold is modified. This feedback loop allows the system to continuously refine its sensitivity to individual user characteristics and environmental conditions, maintaining optimal performance even as usage patterns evolve.
9. The method according to claim 6 , wherein at least one of said first, second, and third thresholds depends on an orientation of the mobile device.
In the gesture recognition method, at least one of the symmetry, energy, and dominant frequency thresholds depends on the current orientation of the mobile device. For instance, the threshold for downward flicks might be different if the phone is held vertically versus horizontally. This accounts for the fact that the same gesture might produce different sensor readings depending on the device's spatial position.
10. The method according to claim 6 , wherein said first, second, and third thresholds depend on the axis.
In the gesture recognition method, the symmetry, energy, and dominant frequency thresholds depend on the specific sensor axis being measured. For example, the thresholds for X-axis movement might be different from those for Y-axis or Z-axis movement. This accounts for the fact that gestures might exhibit different characteristics along different axes, allowing for more accurate and nuanced recognition.
11. The method according to claim 1 , further including a step of automatic adaptation of a size of the time window to durations of gestures in the set of predetermined gestures or to durations of the user's gestures.
The gesture recognition method automatically adapts the size of the "time window" (the duration over which sensor signals are analyzed) based on the expected duration of gestures in the predetermined gesture set, or the actual durations of the user's gestures. This dynamic adjustment ensures that the system captures the complete motion of each gesture, improving recognition accuracy by aligning the analysis window with the temporal characteristics of the gestures being performed.
12. The method according to claim 1 , wherein said axis is a measurement axis of said sensor.
In the gesture recognition method, the "axis" along which symmetry and energy are calculated is a measurement axis of the motion sensor itself (e.g., the X, Y, or Z axis of an accelerometer or gyroscope). This constraint simplifies the analysis by directly utilizing the raw sensor data without requiring complex transformations or projections.
13. The method according to claim 1 , wherein said substep of calculating the value of the parameter indicative of the energy of the candidate gesture along at least one axis is performed when said substep of calculating the value of the parameter indicative of the symmetry of the candidate gesture with respect to said axis indicates a symmetry.
In the gesture recognition method, calculation of the "energy" parameter is only performed if the "symmetry" parameter indicates that the gesture exhibits a degree of symmetry. This optimization avoids unnecessary energy calculations for gestures that are clearly non-symmetrical, reducing computational load and improving overall system efficiency.
14. The method according to claim 1 , wherein said time window is a sliding window.
The gesture recognition method uses a "sliding window" approach. This means that the time window for analyzing sensor data moves continuously over time, allowing the system to detect gestures that may start at any point in time. As new sensor data becomes available, the window shifts, enabling real-time and continuous gesture recognition.
15. The method according to claim 14 , further comprising a step of counting a number of successive time samples for which, over a corresponding time window, a candidate gesture is detected, and a step of comparing a detected candidate gesture with a set of predetermined gestures when said number of successive time samples reaches a predetermined number.
The gesture recognition method counts how many consecutive time samples, within a sliding time window, result in the detection of a candidate gesture. Once this count reaches a predefined number, the method compares the detected candidate gesture with a set of predetermined gestures. This confirmation step requires sustained evidence of a gesture over time before triggering full recognition, reducing false positives and increasing reliability.
16. A recognition device for continuous recognition of gestures of a user of a handheld mobile device comprising a motion sensor assembly, and including at least a module configured to perform the method according to claim 1 .
A device for recognizing gestures on a handheld mobile device with a motion sensor includes a module that performs the following steps: continuously monitors sensor signals over a time window to detect gestures; calculates a "symmetry" value indicating how well the gesture returns the device to its starting position or orientation; calculates an "energy" value representing the gesture's intensity along at least one axis; and compares the detected gesture with a set of stored, known gestures. The module uses these parameters to quantify the shape and strength of potential gestures, helping to identify and classify them.
17. A handheld mobile device comprising a recognition device according to claim 16 .
A handheld mobile device incorporating a gesture recognition system, the gesture recognition system comprising of at least a module that performs the following steps: continuously monitors sensor signals over a time window to detect gestures; calculates a "symmetry" value indicating how well the gesture returns the device to its starting position or orientation; calculates an "energy" value representing the gesture's intensity along at least one axis; and compares the detected gesture with a set of stored, known gestures. The mobile device uses these parameters to quantify the shape and strength of potential gestures, helping to identify and classify them for device control.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
December 2, 2014
May 30, 2017
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.